01. RA - ОПИСАНИЕ : DESCRIPTION

“... 99) Another and not the last goal in RA is to provide

a thoroughly indexed, consistent register of all decisions of all phases of all cycles of creation, development and deployment of a formal system, including its documentation.”

The subject of this article is a description of the purpose, principles of construction and prospects of a certain solution, grouped under the name - RA:(Relate Assemblies - Relink Assemblies).

The context of this article applies simultaneously to many sections of information: technical documentation, mathematics, philosophy, history. Such a broad scope is justified by the main, but not explicit, purpose of the RA solution (which manifests itself as a consequence of its capabilities and prospects).

The technical purpose of the RA solution is a tool, a set of resources and mechanisms that provide the ability to describe other technical solutions, i.e. RA is just another description tool. The ways in which RA is proposed to describe something refer to it most likely another formal decision language of the type: C, Basic, JavaScript, … Assembler. The ellipsis in the enumeration, at the same time indicates the claim to combine in RA all the existing features of all useful (my favorite:) formal languages, but this is just another property of it.

The approaches and methods embedded in RA, which ensure its purpose, have an effective ability to infinitely improve the solutions described with the help of RA. Other methods (formal languages, solutions,...) also do not prohibit improvement, but do not have effective mechanisms for continuous improvement. Even a regular text editor, in theory, allows you to create a perfect text describing a certain perfect solution. BUT it will not help you in detecting and correcting logical errors.

The effectiveness of RA mechanisms lies in the ability to automatically detect contradictions and guarantee their elimination. In fact, RA is based on the most rigorous principles of relational set algebra, which are usually implemented in relational database management systems (RDBMS).

To the well-known restrictions, designated as ACID compliance (wiki), are added another five or more rules and restrictions. And the paradox is that despite the continuous restrictions, the RA user gets to the paradise of total logical freedom, where you are free to make mistakes as much as you want, but helpful and tireless assistants are always ready to help correct everything and analyze the effectiveness.

Any modern implementation of RDBMS (MS-SQL, MySQL, Postgre ...) allows you to create consistent descriptions of something (why do we need another one?). The problem is that in well-known RDBMS, you can either turn on the mechanisms for tracking contradictions and immediately create something useful and not contradictory, but simple, or turn off those mechanisms and describe something smarter than 2+2, but most likely containing contradictory expressions, and unfortunately, if you can still put up with error detection methods, then with corrections, especially automatic ones, everything is much worse.

The paradox in RA is that with all the mechanisms for prohibiting contradictions enabled and reinforced by its own prohibitions in RA, you are allowed to create contradictory statements and gradually correct and supplement them. And all this because RA, how should I say it, loves and cares for mistakes, i.e. they are not mistakes at all.

It turns out that RA is focused on creating and managing errors? It's true. But the harsh truth, according to the author, is that there is no single solution to any problem, absolutely unmistakable and will never appear. RA also helps to approach the solution with any predetermined degree of infallibility.

The mathematical foundations of RA - as already mentioned, lie in the field of relational set algebra, which actually relies on: ”General set theory tightened by the Zermelo/Frenkel axiom of choice " ZFC-ST: (ZFC Set Theory wiki). 

The cruelty of the axiom of choice was revealed as a suggestion from Messrs. Z/F to resolve the well-known contradiction of the so-called “Naive Cantor Set Theory” that appears when trying to describe “The set of all ‘ordinary’ sets”, which, among other names, is known as the Russell paradox (wiki). 

This name of the paradox I chose not by chance, because Russell offered his solution to the paradox revealed in his "Theory of Types”. But something did not stick together. And I guess that and as it turned out, my approach, in fact, continues his approach.

Even very long ago, and very often, before and after Russell, attempts were made to describe a consistent hierarchy of all entities - categories - classes, and equally unsuccessful. Bertrand, admiring the prospects of set theory and the success in formal descriptions of Piano, rightly suggested that if all the elements of any sets are ordered according to an ideal hierarchy of types described in the context of a formal system, then the paradox is removed by a trivial procedure.

The problem, however, was to create a perfect type hierarchy. Russell created and refined it all his life, and in the end even renounced his views. Perhaps under the pressure of the success of the application of ZFC-ST or because of the face-melting theorems of Godel about the incompleteness of formal systems (wiki).

I, too, constantly did not like something in the approaches of ZFC-ST, and Godel generally angered me like a red rag for a bull. What did I like? I found out right after reading Descartes with his research methods.

The fact is that Rene actually rebelled against the dominance of scholastics and scientific pessimists (with their millennial disputes about where stick began) and in his research revived the importance of systems thinking and experience based on rationality and Aristotelian logic.

So Godel, it seems, is not against logic (he proved his theorems strictly mathematically)? But in his interpretation of the applicability of the “incompleteness " theorems, he actually revived the scientific pessimism that Descartes had so successfully fought.

Recall that in the 1st theorem, Godel proved the possibility of the existence of fundamentally unprovable statements made strictly according to the rules of any formal system. And in the 2nd, he hammered the last nail in the hope of someday to create a system that will at least prove its consistency, all the more so to prove anything else.

Although, I agree with the claims that it was the ZFC-ST approach that actually helped to develop formally accurate whole sections of modern mathematics. And Goedel's theorems, significantly saved research effort-helping to understand quds and how to dig deep. Yet the coffin of hopes with the nails from Godel remains in its place unmoved.And how can RA help here?

Let's start by analyzing what is wrong with ZFC-ST? Then we will get to Godel. The fact is that the axiom of choice significantly narrows the range of possible applications of Cantor's "naive" set theory.

Georg noticed that any reasoning can be represented in the form of some elements of certain sets, which, together with the simple rules of working on sets, provides a basis for abstracting from the subject of reasoning, and therefore gives hope to formally describe and prove any formal systems.

Unfortunately, the axiom of choosing ZFC-ST, in fact, excludes all non-ideal systems from the list of any systems, i.e. the most valuable real examples are excluded.

It is interesting to note here that Godel's theorems, for example, in the form of RDBMS with ACID constraints always enabled, do not apply. Because the proof is based on the statement that for any formal system it is possible to construct self-referential expressions (including a reference to oneself as one of the elements of the expression), which lead to contradictions and the inability to meaningfully evaluate such statements.

So, in RDBMS with ACID (Enforced ON), there is no way to build self-referential expressions. They are prohibited by the system mechanisms at the time of the attempt to enter them.

In fact, here we are faced with the incompleteness of the proofs of the “incompleteness” theorems. Godel reformulated the Russell paradox strictly mathematically and there deduced the statement that it must appear in all formal systems, but the example above refutes this. There are more examples, see below APTS example of a formal system (p.30 below).

The Procedure for Removing Contradictions of Russell/Godel type (PRC-RG). Imagine an RDBMS with ACID (Enforced ON), filled with data about the Ideal Hierarchy of All Types by Russell (IHAT-R). Such a data demon (a closed and independent set of data together with mechanisms for manipulating them. The term from the OOP wiki) will be able to:

parse any external (input to the system) contradictory statements by forcibly breaking one of the links that: a) form a ring of self-reference and exactly the one that: b) contradicts the relationships in the ideal type hierarchy IHAT-R.

A broken connection is not forgotten forever, but is marked as secondary and counted last. For the case when we have one contradictory statement and there is a criterion for excluding the contradiction IHAT-R, we can decide that this statement ceases to be contradictory.

For cases where there are a lot of such statements, we simply convert them all to “correct" in the order of the dynamically calculated rank of the statement, which is defined as the maximum rank of all elements of the statement + 1. The rank of the elements is determined according to the rank of the type of this element in the IHAT-R.

It can be assumed that in the context of another example of an ideal ZFC-ST filled with All Possible and only True Statements (APTS), Goedel's theorems lose their meaning altogether. Ah! If we had APTS or at least  IHAT-R.

But what if instead we build mechanisms with a sufficient initial number of perfectly ranked types. And, if we are lucky, among the first registered types there will be many from the top of the IHAT-R hierarchy. And to all RDBMS with ACID (Enforced ON) mechanisms, we will add the procedures described above (+/- a couple more).

Such a data demon would potentially be ready to break Godel's “incompleteness" theorems and fulfill the dreams and aspirations of Georg and, after him, Bertrand. It (the solution) will be infinitely close to the IHAT-R, which means that it will solve the issues of contradictions with sufficient confidence and at the same time without restrictions, i.e. for all, and not only ZFC-ST cases. Now we have two new questions. What is "lucky" in the context of building the above plan? At what point and who will decide that the demon is already matured for victory?

Luck, for example, in preference, I always managed to replace with the combination: brains + diligence. And in this case, for 20 years, with the tenacity of a bull, I repeatedly began to build and reached internal insurmountable dead ends. Threw and after a while again began to build the similarity of IHAT-R + some number of procedures. Someone will say - and that's it! Well, add here a pinch of luck, even Godel does not forbid it. No, really? Read on and find out.

A little more history (as promised in the introduction), the most ridiculous and discouraging cases were when you at some stage, notice that your last correction-well, very necessary, a couple of cycles ago (read years) was canceled by you, well, for very strict reasons. Still, at some point, the backbone of unchanging ideas was formed, but the absolutely unbearable trampling around seemingly small things continued. The story is long, I'll tell it some time later, and here I also promised about philosophy.

Maybe it's time to talk about achievements and how they are offered to receive them? Well, no, if someone did not understand, and about sets and about philosophy-it's all about what RA is and what and how it can do it. Russell spent his life trying to convince himself that what he had planned was useless or unrealizable. I have spent almost 20 years (in fact, I have been going to this decision all my life), and you are afraid to spend 20 hours (or days)?

About philosophy in RA. Strict answers to the questions asked are beyond my and RA's current capabilities. But mathematics + a number of philosophical and logical principles underlying the construction of RA give in the sum-er-er, what we have as a result:)

The logical basis of RA stands on one law and one principle, they act as axioms. Everything else can be deduced as natural consequences purely logically and without going beyond RA. The 1st law, which is the only law external to RA , is the Law of the excluded 3rd state. Recall that it says:

Any statement can be evaluated as one of 2: true or false, and the 3rd is not given. There it is. The law is the 1st, uses the number 2 in the content, and in the name contains 3:) Miracles! We will talk a lot more about the miracles with numbers in RA.

The 1st law about the excluded 3rd state has an important consequence, which actually controls the most important mechanism that determines the approach to solving the possibilities stated in the RA. In RA, the presumption of the principle of orderliness is realized and is not otherwise consistent with the criterion of Causal Relationships (CR), which are the natural consequences of the postulation of the 1st law as the main one.

What does the presumption mean here? This means that everything that is stored in RA is always either strictly ordered according to the CR criterion or can be unambiguously brought to order in one pass of the sorting procedure.

Moreover, here we come across the first and very important addition to the ACID set of constraints. Everything that is stored in RA is not just sorted by the CR criterion, but the most important thing is that the primary key (the numeric index of each storage element) exactly matches the number of the element in this queue.

At this point, the Founding Fathers of Relational Algebra (FFRA) must have fallen into a hiccup. The fact is that the FFRA are convinced that is very important and in no case can we assume a specific position of the elements in the set. Or use the item position information in some useful way in the future.

This is a saying: let the computers / engines decide. They say that it is not a royal business to pick at numbers. But the most important thing is that this supposedly solves a very important degree of invariance of the elements of sets to their positions.

Objections to the FFRA arguments. Invariance does appear, but is it really that important? And what is the price of this, it is not clear why this is necessary invariance. I'm going a little overboard here.

Indeed, it is known from the history of other developments/research that any additional degree of invariance brought by developers to the foundations of systems is useful. But the question of price is not removed.

The ordinal numbers of the elements of the set or their other numerical designations are associated with a lot of important mechanisms for manipulating sets. Effective procedures are implemented for sorting, searching, combining, ... Moreover, they work even more effectively if you use unique pseudo-random numbers instead of ordinal numbers. This fact is a strong argument in favor of the FFRA position.

But what if I show how, in a concrete example of a complex system of interconnected sets, the search for elements is generally replaced by a simple arithmetic expression for calculating the position of an element, just because of the ordinal numbers of these elements.

And for all other operations on sets, there are no less effective algorithms that use ordinal numbers. But the most important argument in favor of ordinal numbers, namely, in accordance with the permanent sorting by the CR criterion, is the method of removing logical contradictions and related restrictions in ZFC-ST, set out in the PRC-RG above.

The Principle of Maximum Normalization (PMN) or another nail in the position of FFRA. The FFRA themselves spend a lot of time explaining the importance of the highest possible level of Normalization of Relational Databases (NRDB) in their algebra. 

What is the level of NRDB, I do not have time to disclose here (google it and you will be rewarded wiki). Here we just mention that the higher it is, the supposedly more useful for logicians/mathematicians (in short, botanists), but not programmers.

It is mentioned that the 5th level is extremely smart, there is a 6th level behind them, but even the 5th is considered purely theoretical and even FFRA is not recommended in practice. Like smart it is smart, but there are also a lot of overhead costs (wow, they also care about the price).

And what if I show by the example of the working system that it makes sense to introduce another level of higher and essentially marginal normalisation? Recall that in the 5th level, we are talking about the fact that in a closed system of linked relations there should not be a single repeat (each element is present only in one set) and there should not be empty links (if you need to specify a link with the “parent” element in the definition of an element, then divide your set into a subsets until there are no elements with empty links in any of them).

Now imagine one relation (in relational algebra, sets are called relations - roughly speaking, this is a table), in which there are only 3 essential descriptors of the set element. (1) - the sequence number according to the CR criterion. (2) - the content/value of the element and (3) - the context/relationship with the parent element - the number of the parent element in this set.

Such a set can be effectively maintained in a non-contradictory order only if it is always sorted by the CR criterion and the parent links lead only in one direction and precisely to the top (to the earlier elements - smaller numbers).

It turns out that the ordinal number gives us another higher level of normalization, and you sang something to us about the invariance of the order number. Well, what's the use of the new normalisation? Just for a second - we have just derived the structure of a universal set, such that it is possible to replace all the structures of all sets, and therefore we have obtained invariance to the structure of the set. What is more valuable? Time will show.

Just as the 5th level of the NRDB is inconvenient, the PMN proved to be even more difficult in the process of initializing it. Usually, any person does not immediately see the contradictions. It is not sufficiently painstaking and attentive when checking and correcting. He's impulsive-nothing-nothing-little-little-and then bam.

All this made me start over 100 times. In addition, as for high-level NRDB, the efficiency and convenience of manipulating data in the PMN is even worse. Abandon the PMN? The prospect of losing universality didn't bother me much, but to abandon the prospect of automatically correcting contradictions was to abandon the whole idea. The idea of combining the advantages of low levels of NRDB with PMN proved to be a lifesaver. How?

PMN-4-NRBD: Maximum normalization by means of the usual normalization of the 4th level. Something here already confusing? Everything falls into place if we assume that all levels of normalization without a unique index for all entities are normalized in the first dimension, while systems that provide pass-through uniqueness are proposed to be normalized separately in the second dimension.

Thus, all previous levels of normalization are denoted as: 0:0, 0:1, ... 0:5, 0:6, and the new ones as 1:0, 1:1, ... . Since, for our taste, the most favorable level of normal normalization is - 0:4, our target level will be-1:4.

As a result, we are forbidden to repeat elements, but empty links are allowed, if there is a pass-through unique index for all sets. The enabling of empty references, as will be shown later, is extremely important in view of other problems of human nature.

The Principle of Organized Disorder (POD). We already know that a person is stupid, lazy and impulsive, but he is also weak. It is known that a person is not able to keep more than 7 alternatives in the focus of attention at the same time. According to my observations, no more than 3.

This implies the need to help him look at local groupings of elements, even to the detriment of the main principle of cause-and-effect relationships (CR). That is, when organizing data, it is necessary to have other criteria besides CR. Moreover, these criteria should be monitored with all the strictness of ACID (Enforced ON), but leaving priority to the CR connections.

Thus, it becomes necessary to divide one relationship about the parent context into different categories of relationships and their priorities for consideration. Moreover, from the point of view of the principle of the minimum necessary set of descriptors, it would always be possible to do so with a single connection. The division is made to please the weaknesses of Homo-simplicity-lazy-emotional and weakly.

So we have connections (Near Context) for CR, (Far Context) for local and other logical groupings, (Holding Type) - features of storing and interpretation, (Naming Type) - features of designations well, that's enough for now. Moreover, only the Near Context link should not be empty (except for the top of the hierarchy), and the ACID rule (Enforced ON/Cascade Change=ON/Cascade Delete=ON) should apply to the CR.

The rest of the links belong to the POD category with ACID rules (Enforced ON/Cascade Change=ON/Cascade Delete=OFF or (Set NULL)). That is, for all content changes, for all types of links, the changes are cascaded in all child applications, but the deletion is cascaded only for the CR.

For the rest, it is possible and often necessary to set an empty relationship without deleting the child elements of the set. A person does not read long data chains well -> it is necessary to group the data usefully -> hence empty connections appear, moreover, they are needed. Thus, we have just proved that for systems with human participation, the 4th level of normalization should be the last in all dimensions.

(Far Context), introduced for logical groupings to aid human weaknesses, later proved to be an extremely powerful broad-purpose description tool. In the 1st, it allows you to store Russell/Godel type contradictions in RA under full control. How?

(Far Context) allows you to create self-reference links, but the main ones are the links (Near Context): (CR) that exclude contradictions. Thus, specially marked cases of contradictions formed in the (Far Context) contour of hierarchies allow them to be handled correctly: to bypass, transform, ... .

It is this feature that makes RA a non-limited system, unlike systems subordinate to ZFC-Set Theory, and implements the stated BOD above. Other notable features (Far Context) and how it implements many assignments at the same time, see below in section 1.3. Important distinguishing features in RA.

The Principle if You Can Not but really Want to then You Can (PYCNWYC). In modern RDBMS, much of what we need is either forbidden (MS-SQL) or works for simple cases, and then blunts (MS-Access). Let me remind you that we need to organize many connections from one set to another one specific set.

The nature of these restrictions is clear. After all, the founding fathers FFRA do not need ordered sets. And although we are ready to take an oath not to violate the principle of ordering by CR, there are no such options in the RDBMS settings. Creating your own RDBMS is not a forbidden path, but it is very expensive. It remains to expand the standard features.

Here we will be very helpful to the already mentioned ACID rules, Transactional Integrity (TI) rules and Data Change Triggers (DCT), which are built into RDBMS. With the help of DCT, you can, exactly always, run some data processing procedures, immediately after any changes in the data (or exactly instead of changes).

The rules of TI guarantee that despite possible delays or failures in the system, a certain set of data processing rules designed as a single transaction will either work out guaranteed to the end, or completely will be canceled, exactly to the state before the start of the transaction.

If you use the laws of logic, DCT and TI at the same time, then there is a chance to expand and/or impose new useful restrictions for RDBMS. We can use the ACID(Enforced OFF) option to mark the links we need, and all that RDBMS does for ACID(Enforced ON/Cascade Change=ON/Cascade Delete=(Set NULL)), do it yourself using the DCT and TI.

It was said and done, but it was not there. The necessary processing procedures are trivial, but it turned out to be a lot of small things and not a trivial cascade of recursions and they are durable in the work, and for the DCT there is an implicit, but annoying requirement to quickly work everything out.

You can persistently design more and more effective solutions, but here is where (this is amazing) the seemingly negative quality of a person helps him out - Mother Laziness. The solution on the other side of the connection turned out to be easier, although with its drawbacks.

You can artificially keep many copies of the main (parent) set up-to-date, one for each association with ACID (Enforced ON). Copies are kept up-to-date using the same DTC+TI. The solution is redundant and seems to blatantly violate the principle of storage uniqueness. Not really.

Responsible uniqueness is provided in one copy, the rest play the auxiliary role of shadows. Thanks to DTC+TI, we get another data demon, about which we can assume that it's like just one copy. Storage costs remain, but this solution will work as a temporary compromise. What if the main hopes of the RA will be untenable?

The main hopes of RA. So. The potential for possible versatility of RA may be the main useful property in the future, but at the moment it is least interesting.

RA is a potentially non-contradictory formal system, within which it is possible to formally prove its consistency, as well as to make meaningful conclusions about the inconsistency of any other formal system without restrictions imposed by the system: Set theory with the axiom of choice (ZFC Set Theory) and without internal contradictions of the Russell/Goedel type of: Cantor's Naive Set Theory.

Recall the chain of reasoning that leads to such conclusions

RA: { RDBMS: {ACID (Enforced=ON), ... } => example of a formal system without Russell/Goedel contradictions

+ Data Domains: { Documents[]+Objects[]+Types[]+...} with Global Unical Index through all data domains with

ACID: {(Enforced=ON) + (Cascade Change=ON) + (Cascade Delete=ON) } for the main context relationship named (Near Context): (CR): (Causal Relationships)

+ SubSet (PHAT) => about 400 elements from (IHAT-R): (Ideal Hierarchy of All Types by Russell)

+ (PRC-RG): (Procedure for Removing Contradictions of the Russell/Godel type) = > (1) take the following contradictory expression. (2) find the cycle of links leading to self-reference. (3) find the link that contradicts the direction of the links in the Partial Hierarchy of All Types (PHAT). (4) mark it as an auxiliary link (not a CR). (5) catalog the resulting expression in the PHAT.

}

Remark. The stated chain of reasoning only outlines a rigorous proof, but cannot be considered as such.

How are Multiple Assignments (MA) implemented in the secondary contour of the hierarchy in the field (Far Context)? A Pass-through Single Primary Index (PSPI) for all data domains allows you to implement a number of exotic storage formats with all the strictness of ACID (Enforced=ON). For example,

Context Expressions (CE) as independent entities stored in a separate data domain, each with its own unique PSPI.

CE (type i): {(PSPI) + (Far Context) [0] + (Far Context)[1] + ... + (Far Context)[n]}, where n is the number of elementary contexts in CE, and (type i) describes the purpose of each elementary context.

Thus, first, the desired MA is described as a new element in the CE, and then the proper PSPI index of this CE is used to describe the target element of the set in its field (Far Context). And everything is strictly in accordance with ACID (Enforced=ON).

CE, in particular, can be used as a description of the sum of tags for elements of sets. The tag description method has proved to be a very convenient mechanism in many formal systems, but in RA it is implemented flawlessly strictly in accordance with the principles of relational algebra, which allows you to create not only lists of tags, but also non-trivial hierarchies of them, with any subsequent refinements without contradictions.

What's in my name for you? In the RA system, in parallel to the development of the structure, the subsystem of object designations developed. Everything for which it is useful to have a name or a symbolic designation is not named according to one of the well-known naming systems today. It is forbidden to form arbitrary combinations of characters for the purpose of object names.

But you can register a textual explanation of the object's purpose and features, which, together with the indication of its type, contexts, and template rules, will automatically generate the object's designation. The text explanations themselves are still indicated in free form in English. In the future, text expressions will also become relational strict, relying on lexemes and lexical expressions of the CE type.

In RA, it is Guaranteed that Changes Propagation in the Designations (GCPD) of objects and that they are accurately rolled back to the previous state in cases of designation conflicts. Naturally, the cases of conflicts in the main data domains are excluded at the time of the attempt to determine them. Conflicts are possible in numerous derived designations.

The core of the RA components is designated according to "contextCONTENT” The Outline Style (cxCN_OS) with the possibility of forming cascading hierarchy designations (the “_"symbol), using special designations for type indications (if such is useful, the “__" symbol).

The designation of faceted classification is widely used, which, in combination with the ability to adjust the location of objects through (Far Context), may violate the sequences formed in view of the CR links, but in accordance with the plan described in the documentation, which is also a component of RA, which can always change and change the designation and location of objects. Oil oiled, but what to do if that's what you need. Therefore, this is the name of the system: Relate Assemblies or Assemble Relationships.

Other styles are also not prohibited, which, in combination with the synonymy mechanisms, allows each developer/user to see objects in the form of familiar designations, which, however, can make it difficult to understand each other.

Effective mechanisms for combining sets in RA allow you to build any Distributed System Management (DSM) schemes without additional descriptions (as is done in the description of RDBMS replicas). Create and maintain complex multi-module data schemas, following the example of multi-modularity of program code. And this is also a consequence of permanent ordering according to the CR criterion.

In addition to the exotic CE (or because of its consequences), it becomes possible to create not just "named enumerations", but any complex Named Structures, Bits, Masks, Codes (NSBMC) and their combinations with the automatic generation of symbolic designations and automatic propagation of changes in both values and designation.

A little magic in the designations of objects of the CE type and we get a new storage format: the Fractal Matrix (FM). This is an object in which the variable dimension of the data matrix is encapsulated in its methods, properties, and designation, which eliminates the redundancy of storage characteristic of the 4th level of normalization of relational databases (NRDB, see above).

Meta Programming Methods (MPM) and mechanisms for creating and applying macros are not just available in RA, but are provided with all the rigor of object programming standards and relational integrity.

The pass-through unified index PSPI of elements allows the creation of Object Logging (OL) for the purposes of effective and constantly available analytics, monitoring, and dynamic macros for testing, training, automation, and for saving and restoring system states.

Effective mechanisms for automatic archiving, multi-level and detailed profiling are seen in the future. Simultaneous active and consistent execution of many versions of the system.

Another and not the last goal in RA is to provide a thoroughly indexed, consistent register of all decisions of all phases of all cycles of creation, development and deployment of a formal system, including its documentation.